Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The antimetabolite 6-mercaptopurine (6-MP) has been widely used as a chemotherapeutic agent and, to a lesser extent, as an immunosuppressant for almost five decades (Elion, 1989). The mechanism of the drug's biological activity is complex and may involve modulation of cellular metabolism in several ways (Tidd and Paterson, 1974; Van Scoik et al., 1985; Martin, 1987). First, 6-MP can be metabolized to nucleotides that are incorporated into nucleic acids; this pathway is believed to be central to the cytotoxic activity of 6-MP. Second, 6-MP can act as a pseudofeedback inhibitor of de novo purine nucleotide synthesis. Finally, by competing with normal purines, 6-MP can form an inhibitory analog-enzyme complex, thus disrupting the synthesis of necessary intermediates and the interconversion of purine nucleotides. Severe bone marrow and liver toxicity associated with 6-MP treatment have led to the design and synthesis of prodrugs that might decrease the systemic toxicity of the drug and ensure its selective delivery to the target tissue. In this regard, a prodrug of 6-MP is defined as a pharmacologically inactive derivative of 6-MP that can be metabolized to yield 6-MP.

Numerous potential 6-MP prodrugs have been examined (Drawbaugh et al., 1976; Nelson and Vidale, 1986; Waranis and Sloan, 1988; Daniel et al., 1989). For example, the 6-MP prodrug azathioprine (Fig. 1A) has been shown to be superior to 6-MP as an immunosuppressant, whereas its therapeutic index against leukemia and adenocarcinoma is similar to that of 6-MP (Elion, 1989). Azathioprine has now replaced 6-MP in the treatment of organ transplant patients (Van Scoik et al., 1985). Hwang and Elfarra (1989, 1991) selectively targeted 6-MP to the rat kidney using S-(9H-purin-6-yl)-L-cysteine (Fig. 1B). The concentration of 6-MP and its metabolites were nearly 2.3- and 90-fold higher in the kidney than in the liver and plasma, respectively. The S-(9H-purin-6-yl)-L-cysteine analogs, S-(9H-purin-6-yl)-N-acetyl-L-cysteine, S-(9H-purin-6-yl)glutathione and S-(9H-purin-6-yl)-L-homocysteine were also shown to selectively deliver 6-MP to the kidney (Elfarra and Hwang, 1993; Hwang and Elfarra, 1993). Similarly, S-(9H-guanin-6-yl)-L-cysteine was shown to selectively deliver the anticancer drug 6-thioguanine to the kidney (Elfarra et al., 1995).

View larger version (11K): [in this window] [in a new window]   Fig. 1.   The structures of three 6-MP prodrugs. A, 6-[(1-methyl-4-nitro-5-imidazolyl)thio]purine (azathioprine); B, S-(9H-purin-6-yl)-L-cysteine; C, cis-3-(9H-purin-6-ylthio)acrylic acid (PTA).

The reported antimicrobial activity of the azathioprine structural analog cis-3-(9H-purin-6-ylthio)acrylic acid (PTA; Fig. 1 C; Turbanova et al., 1981), led us to hypothesize that the biological activity of PTA was caused by its glutathione (GSH)-mediated metabolism to 6-MP, thus making it a potential prodrug of 6-MP. The conjugation of the intracellular tripeptide GSH to azathioprine and the 6-MP prodrug chloropurine is believed to play a key role in the metabolism of these compounds (de Miranda et al., 1973; Hwang and Elfarra, 1993). The conjugation of GSH to electrophilic agents can occur nonenzymatically for some compounds; however, glutathione S-transferases (GSTs) enhance the rates of several reactions. Five isoforms of GST, expressed in a tissue-specific manner, have been characterized in mammals. Four of them; the , µ, , and , are found in cytosol, whereas the fifth is found in the liver endoplasmic reticulum (Morgenstern and DePierre, 1987; Gulick and Fahl, 1995; Hayes and Pulford, 1995). To examine the potential utility of PTA as a prodrug of 6-MP, the in vitro metabolism of PTA to 6-MP was examined in rat liver and kidney homogenate and the roles of GSH and GSTs were characterized. Furthermore, preliminary experiments were carried out to assess the acute toxicity of PTA in rats and to determine whether 6-MP is an in vivo metabolite of PTA. The results presented clearly show that PTA is a prodrug of 6-MP. A preliminary report of this study has been previously presented (Gunnarsdottir and Elfarra, 1999).

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Chemicals. Acivicin, aminooxyacetic acid (AOAA), allopurinol, thioxanthine, GSH, rat liver GST, and xanthine oxidase from buttermilk, were purchased from Sigma Chemical Co. (St. Louis, MO). Human recombinant GST A1-1, GST M1-1, and GST P1-1, expressed in Escherichia coli, were purchased from PanVera Corporation (Madison, WI). Propiolic acid, sodium methoxide, anhydrous methanol, phosphoric acid (85%, w/v) and 6-MP monohydrate were purchased from Aldrich Chemical Co. (Milwaukee, WI). HPLC-grade acetonitrile was purchased from EM Science (Gibbstown, NJ). All other chemicals were of the highest grade commercially available. ¹H-NMR spectra were obtained at the National Magnetic Resonance Facility at Madison (Madison, WI), with a 500 MHz Bruker spectrophotometer using D₂O/NaOD as solvent. Chemical shifts are reported in parts per million, using 3-(trimethylsilyl)-tetradeutero sodium propionate as internal standard.

Synthesis and Characterization of PTA. PTA was synthesized according to the method of Turbanova et al. (1981). In brief, 6-MP (1.2 mmol) was added to 8 ml of anhydrous methanol. Sodium methoxide (approximately 4.4 mmol) was added with continuous stirring until the 6-MP was dissolved, followed by the addition of propiolic acid (1.2 mmol). The reaction mixture was refluxed with continuous stirring overnight (15-18 h) and was quenched by the addition of 4 ml of water. PTA was recovered from the reaction mixture as a white precipitate that formed after the addition of 1 M HCl. HCl was added until no more precipitate was formed. The precipitate was collected by filtration, redissolved by the addition of 1 M NaOH, and reprecipitated by the addition of 1 M HCl. The yield of PTA was between 55 and 79% and purity, as determined by HPLC, was greater than 97%. Melting point of PTA: 235°C (decomposition); reported: 255°C (decomposition; Turbanova et al., 1981). ¹H NMR: 6.30 ppm (1H, d, J = 10.0 Hz, vinyl proton), 7.99 ppm (1H, d, J = 10.0 Hz, vinyl proton), 8.23 ppm (1H, s, purine ring proton), 8.58 ppm (1H, s, purine ring proton).

Characterization of S-(9H-Purin-6-yl)glutathione (PG), a Major Reaction Product of PTA and GSH. A major product formed in reactions of PTA and GSH in buffer only, and in the presence of rat liver homogenate, was collected as described below in HPLC Analyses. The product was desalted by elution through PrepSep C₁₈ extraction column (Fisher Scientific, Fair Lawn, NJ) using 10% (v/v) acetonitrile in water and the eluate was then lyophilized. A UV spectrum of the product, dissolved in HPLC mobile phase at pH 3.5, was obtained using a Beckman DU-7 spectrophotometer. UV: _max 288 nm, _min 243 nm. ¹H NMR: 2.04 ppm (2H, m, glu- protons), 2.42 ppm (2H, m, glu- protons), 3.64 ppm (1H, dd, J = 14.4, 8.4 Hz, cys- proton), 3.73 ppm (1H, t, J = 6.3 Hz, glu- proton), 3.89 ppm (2H, s, gly- proton), 4.02 ppm (1H, dd, J = 14.4, 4.8 Hz, cys- proton), 4.87 ppm (1H, dd, J = 8.3, 4.9 Hz, cys- proton), 8.40 ppm (1H, s, purine ring proton), and 8.70 ppm (1H, s, purine ring proton). Both UV and NMR spectra and HPLC retention time matched those of PG, a compound formed by the reaction of 6-chloropurine and GSH, as previously characterized by Hwang and Elfarra (1991, 1993).

In Vitro Experiments. Tissue homogenate was prepared from male Sprague-Dawley rats (150-215 g; Sasco Laboratory, Omaha, NE). Rats were sacrificed by decapitation, the liver and kidneys were removed and rinsed in buffer (0.1 M phosphate, 0.1 M KCl and 5 mM EDTA at pH 7.4; this buffer was used in all preparations and experiments), patted dry, and homogenized in 3 ml of buffer/g of tissue. The resulting homogenate was used in in vitro assays. To assess the nonenzymatic formation of metabolites, assays using buffer only were carried out in parallel to the enzymatic assays. Because both liver and kidney contain xanthine oxidase, which metabolizes 6-MP to thioxanthine and thiouric acid (Bergmann and Ungar, 1960), all assays, except when purified rat liver or recombinant human GSTs were used, were carried out in the presence of the xanthine oxidase inhibitor allopurinol (5 mM). The 5 mM GSH assay concentration used in these experiments was chosen to represent the intracellular GSH concentration of 3 and 10 mM in kidney and liver, respectively (Sies et al., 1983; Tew, 1994).

In Vivo Experiments. Male Sprague-Dawley rats (155-235 g; Sasco Laboratory, Omaha, NE), were housed individually in plastic metabolic cages (Nalgene Co., Rochester, NY) and allowed food (Purina Labchow, St. Louis, MO) and water ad libitum. The rats were injected i.p. with either PTA (100 mg/kg) in buffer or buffer alone. The rats were sacrificed by decapitation 24 h after treatment, the blood collected in ice-cold test tubes and centrifuged at 1500g for 20 min to separate the serum. Urine from each rat was collected for at least 12 h before and at 6, 12, and 24 h after treatment and analyzed for 6-MP and thiouric acid (TU), an oxidative metabolite of 6-MP formed by xanthine oxidase (Bergmann and Ungar, 1960). Analyses of metabolites in urine were carried out on 1 ml of urine that had been deproteinated by the addition of 75 µl of ice-cold 50% (w/v) TCA and centrifuged at 1500g for 10 min followed by filtration of the supernatant and analysis by HPLC as described below.

HPLC Analyses. The HPLC system used consisted of two Gilson 306 pumps, a Gilson 119 UV/vis detector, and a Gilson 234 autoinjector (Gilson, Middleton, WI). The column used was a Beckman ultrasphere ODS 5 µm reversed-phase C₁₈ (4.6 × 250 mm; Beckman Instruments, Fullerton, CA) with a Brownlee spheri-5 ODS 5 µm (4.6 × 30 mm) guard column (Perkin Elmer, Norwalk, CT), except for the collection of PG, where a semipreparative version of the column above was used. Mobile phase for pump A consisted of 15 mM phosphoric acid in water at pH 3.5 and for pump B 15 mM phosphoric acid in 1:1 acetonitrile-water mixture at pH 3.5. Injection volume was 20 µl and the flow rate was 1 ml/min, except when the semipreparative column was used for PG collection, where the flow rate was 3 ml/min.

1

1

Statistical Analyses. All values are reported as mean ± S.D. with number of experiments (n) as indicated in figure and table legends. Statistical analyses were carried out using Sigma Stat (SPSS Inc., Chicago, IL). Comparison among means was assessed using ANOVA. When significant values were obtained, Fisher's protected least significant difference test was performed to determine which means were significantly different. If the equal variance test failed, Kruskal-Wallis ANOVA on ranks was performed, followed by the Student-Newman-Keuls method to determine which means were significantly different.  was set at 0.05.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

To determine whether PTA was metabolized to 6-MP in vitro, liver or kidney homogenate or buffer only was incubated at 37°C in the presence of PTA (1 mM), GSH (5 mM), and the xanthine oxidase inhibitor allopurinol (5 mM). HPLC analyses of liver homogenate incubations showed two peaks that were not detected in incubations in which PTA was omitted (Fig. 2). The first peak (peak III, Fig. 2D) had a retention time and an ultraviolet absorption spectrum that matched those of reference 6-MP, whereas the retention time, ultraviolet absorption, and NMR spectra of the second peak (peak II, Fig. 2, B and D) were identical with those of reference PG. Both products, 6-MP and PG, were also detected in kidney homogenate and buffer-only incubations, but neither product was observed when PTA was omitted (data not shown). Moreover, the formation of both PG and 6-MP in all incubations was dependent on the presence of GSH. To further verify that the nonenzymatic formation of 6-MP from PTA was indeed GSH dependent but not due to hydrolysis or other breakdown reaction of PTA, PTA was incubated in buffer in the absence of GSH at 37°C at pH 6.4, 7.4, or 8.4 for 5 h. During that time, PTA concentration did not decrease, nor was any 6-MP detected (data not shown). These observations confirm that PTA conversion to 6-MP occurs only in the presence of GSH. Furthermore, PG degradation was not the source of the 6-MP formed nonenzymatically, because PG is extremely stable when incubated in phosphate buffer at pH 7.4 and 37°C (Elfarra and Hwang, 1996).

View larger version (21K): [in this window] [in a new window]   Fig. 2.   Representative HPLC chromatograms of PTA metabolites formed in vitro. A, liver homogenate containing GSH (5 mM) and allopurinol (5 mM) at 288 nm; B, liver homogenate containing PTA (1 mM), GSH (5 mM), and allopurinol (5 mM) at 288 nm; C, liver homogenate containing GSH (5 mM) and allopurinol (5 mM) at 323 nm; D, liver homogenate containing PTA (1 mM), GSH (5 mM), and allopurinol (5 mM) at 323 nm. Peak I, allopurinol; peak II, PG; peak III, 6-MP; peak IV, PTA.

Both metabolites, 6-MP and PG, were formed in a time-dependent manner in liver homogenate or buffer-only incubations containing PTA (1 mM), GSH (5 mM), and allopurinol (5 mM). Time-dependent increase in 6-MP accumulation was also observed when kidney homogenate was used in these assays, but accumulation of PG in kidney homogenate incubations was only observed during the first 15 min. All reactions except the formation of PG in kidney homogenate were linear for at least 2 h, with correlation coefficients greater than 0.98. A comparison of the rate of PG accumulation in buffer and liver homogenate revealed that PG is formed at the same rate in both cases (Table 1). The initial rate (the first 15 min) of PG accumulation in kidney homogenate was, however, much lower than that observed in buffer or liver homogenate, suggesting further metabolism of PG in the kidney homogenate. Analyses of the rate of 6-MP formation in buffer, liver, and kidney homogenate revealed that 6-MP accumulates 2-fold faster in liver homogenate than in buffer, whereas 6-MP accumulation in kidney homogenate is 3.6- and 7-fold faster than in liver homogenate and buffer incubations, respectively. The difference in liver and kidney accumulation of 6-MP was not due to difference in protein concentration in the assays, because the protein concentrations for liver (2.32 ± 0.03 mg) and kidney (2.50 ± 0.13 mg) homogenate incubations were not significantly different. Preliminary experiments also showed that both liver cytosol and microsomes enhanced the formation of 6-MP (data not shown).

PG has been shown previously to be metabolized to 6-MP in rats in vivo and in isolated rat kidney cells by the sequential action of -GT, dipeptidase, and cysteine-conjugate -lyase (Hwang and Elfarra, 1993; Lash et al., 1997). The limited time-dependent accumulation of PG and increased 6-MP formation in kidney homogenate incubations suggest that in kidney homogenate, PTA was converted to PG, which was subsequently metabolized by -GT, dipeptidase, and -lyase to give 6-MP. To provide further evidence for this hypothesis, studies were carried out using the -GT inhibitor acivicin, and the -lyase inhibitor AOAA (Fig. 3). Acivicin (1 mM) increased renal PG accumulation 3.5-fold, whereas 6-MP accumulation decreased to 70% of that observed in buffer-only incubations. Acivicin had no effect on PG or 6-MP accumulation in liver homogenate incubations. AOAA (1 mM) had no effect on PG accumulation in either liver or kidney homogenate or 6-MP accumulation in liver homogenate incubations. However, AOAA decreased 6-MP accumulation in kidney homogenate incubations to 67% of that observed in incubations containing buffer only. Thus, these results, which are in agreement with the previously reported pathway of PG metabolism in rat kidney (Hwang and Elfarra, 1993; Lash et al., 1997), provide strong evidence for the role of -GT, dipeptidase, and -lyase in the renal metabolism of PTA to 6-MP.

View larger version (30K): [in this window] [in a new window]   Fig. 3.   Effect of the -GT inhibitor acivicin and the -lyase inhibitor AOAA on the accumulation of PG and 6-MP in liver and kidney homogenate (n = 3). The nonenzymatic formation of PG and 6-MP has not been subtracted from the results. Homogenate was incubated at 37°C at pH 7.4 in the presence of PTA (1 mM), GSH (5 mM), allopurinol (5 mM), and either acivicin (1 mM) or AOAA (1 mM). The protein concentration in liver and kidney homogenate incubations was 2.32 ± 0.03 and 2.50 ± 0.13 mg, respectively. Reactions were terminated after 30 min and samples analyzed by HPLC as described in Materials and Methods. A, PG; B, 6-MP. *, statistically different (p < .05).

The finding that the rate of 6-MP formation was increased in the presence of liver homogenate compared with buffer-only, whereas formation of PG was unaffected, suggests that GSTs, which are present in high concentration in the liver, metabolize PTA to 6-MP, but are not required for PG formation. To examine further the effect of GST on the rate of PG and 6-MP formation, purified rat liver GSTs were incubated in the presence of PTA (1 mM) and GSH (5 mM). Control incubations containing buffer were run in parallel. A marked increase in 6-MP accumulation was observed in the presence of purified GST compared with buffer-only incubations; the rate of 6-MP formation in the presence of purified rat liver GST was 9.4 pmol/ml/min compared with 5.1 pmol/ml/min in buffer-only incubations. No difference in PG accumulation was observed in incubations containing purified GST compared with buffer-only incubations. Furthermore, no difference in the rate of 6-MP or PG formation was observed between control incubations containing boiled protein and buffer-only incubations, showing that the increased 6-MP formation is not simply a nonspecific consequence due to the presence of protein in the reaction mixture (data not shown). To characterize the relative efficacies of different GST isozymes in catalyzing the formation of 6-MP from PTA, purified human recombinant , µ, and  isozymes (10 U/ml) were incubated in the presence of GSH (5 mM) and PTA (1 mM) (Table 2). All reactions were linear for at least 2.5 h with correlation coefficients greater than 0.96. The rate of 6-MP formation was 1.7-fold higher in the presence of GST A1-1 than in buffer alone, whereas incubations containing GST M1-1 and GST P1-1 increased the rate to 1.3-fold of that observed in buffer. No difference was seen in PG accumulation in incubations containing enzymes or buffer only. These results show that GST can directly catalyze PTA metabolism to 6-MP.

To investigate further the mechanisms of nonenzymatic formation of PG and 6-MP, the effects of pH on the rates of 6-MP and PG formation were determined by incubating PTA (1 mM) and GSH (5 mM) in buffer at 37°C for 1 h at pH 5.4, 6.4, 7.4, 8.4, and 9.4 (Fig. 4). The rates of PG formation at pH 5.4 and 6.4 were not statistically different (approximately 1.2 nmol/ml/min), but the rate increased with increasing pH up to a maximum rate at pH 8.4 (2.44 nmol/ml/min). At pH 9.4, the rate of PG formation decreased significantly. A very different profile was observed when the effect of pH on nonenzymatic formation of 6-MP was examined. The highest rate of 6-MP formation (10.2 pmol/ml/min) was observed at pH 5.4, but the rate then decreased with increasing pH. No statistical difference was observed in the rate of 6-MP formation at pH 8.4 and 9.4, whereas 6-MP formation rate reached its minimum (approximately 4.9 pmol/ml/min). The different pH profiles for the rate of nonenzymatic formation of PG and 6-MP suggest that these products are formed through different mechanisms. In addition, these results are consistent with the previous finding that PG was stable and did not form 6-MP nonenzymatically (Elfarra and Hwang, 1996).

View larger version (16K): [in this window] [in a new window]   Fig. 4.   Effect of pH on the rate of nonenzymatic formation of PG and 6-MP in buffer. PTA (1 mM) and GSH (5 mM) were incubated in buffer at 37°C at pH 5.4 (n = 4), pH 6.4 (n = 4), pH 7.4 (n = 3), pH 8.4 (n = 3), and pH 9.4 (n = 3). Samples were taken every 15 min and analyzed by HPLC as described in Materials and Methods. A, PG; B, 6-MP. Data points with the same superscript are not significantly different (p > .05).

Preliminary studies were carried out to investigate whether PTA is metabolized in vivo to give 6-MP. Four rats were treated i.p. with a PTA dose of 100 mg/kg. HPLC analyses carried out on urine collected from treated rats showed clearly that 6-MP is formed in vivo and excreted in urine. Moreover, TU, a metabolite of 6-MP formed by enzymatic oxidation by xanthine oxidase, was also detected in urine of the treated rats (Fig. 5). At the dose used, a total of 0.5% of the PTA dose was recovered from urine as the metabolites 6-MP and TU. Approximately 85% of both 6-MP and TU was excreted during the first 6 h after treatment, whereas the remainder of the metabolites was recovered in urine collected between 6 and 12 h after treatment. Unmetabolized PTA was also detected in urine, but due to its low solubility in water at low pH, quantitation could not be achieved. Acute toxicity of PTA was assessed by measuring blood urea nitrogen and glucose concentrations, aspartate and alanine aminotransferase activities in serum, and glucose concentration and -glutamyltransferase activity in urine. No difference in these parameters was observed between treated and untreated rats (data not shown).

View larger version (10K): [in this window] [in a new window]   Fig. 5.   Representative HPLC chromatograms of metabolites formed in vivo after rats were administered PTA (100 mg/kg). A, urine collected before treatment; B, urine collected after treatment. Peak I, TU; peak II, 6-MP.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

In this work, we provided evidence for both in vitro and in vivo metabolism of PTA to 6-MP. In vitro, PTA is metabolized in a GSH-dependent manner to yield PG and 6-MP. The different rates of PG and 6-MP formation in liver and kidney homogenate (Table 1), different effects of the inhibitors acivicin and AOAA on PG and 6-MP accumulation in liver and kidney homogenate incubations (Fig. 3), and different pH profiles for the nonenzymatic formation of PG and 6-MP (Fig. 4), indicate that the in vitro metabolism of PTA occurs via two distinct pathways.

The first proposed pathway to 6-MP formation is through formation of PG, the major metabolite of the reaction between PTA and GSH in vitro. Formation of PG appears to be nonenzymatic, as its rate of formation did not increase upon addition of liver homogenate (Table 1) or purified GSTs (Table 2), compared to buffer-only incubations. PG is not formed when GSH is incubated with 6-MP. A possible mechanism that might explain the nonenzymatic formation of PG, which is consistent with the effect of pH on the rate of the reaction, is outlined in Fig. 6A. The sulfhydryl group of GSH in solution has a pK_a value of approximately 8.7 to 9.3 (Jung et al., 1972; Reuben and Bruice, 1976). The increase in rate of PG formation with increasing pH is consistent with increased nucleophilicity of GSH at higher pH due to deprotonation of the sulfhydryl group, which facilitates the nucleophilic attack of GSH on the C-6 carbon of the purine ring. The decrease in the reaction rate between pH 8.4 and 9.4 might be caused by the deprotonation of the N-9 nitrogen of the purine ring; pK_a values between 8.5 and 10 have been reported for the N-9 nitrogen in several 6-substituted purines (Albert, 1971; Mercier et al., 1991). Deprotonation of the N-9 nitrogen results in the formation of an anion, which makes the purine ring much less reactive toward nucleophilic attack.

View larger version (22K): [in this window] [in a new window]   Fig. 6.   Possible mechanisms for the nonenzymatic formation of the PTA metabolites PG and 6-MP. A, PG; B, 6-MP.

PG has previously been shown to be metabolized to 6-MP in the rat kidney, both in vitro and in vivo, by the sequential action of -GT, dipeptidases and cysteine-conjugate -lyase (Hwang and Elfarra 1991, 1993; Lash et al. 1997). Our observations of a limited time-dependent accumulation of PG and increased 6-MP accumulation in kidney homogenate incubations agree well with the reported metabolic pathway of PG to 6-MP. Moreover, the observed increase in PG accumulation and decrease in 6-MP accumulation in kidney homogenate incubations containing the -GT inhibitor acivicin, and the decrease in 6-MP accumulation in kidney homogenate incubations containing the -lyase inhibitor AOAA, provide further support for the proposed biotransformation pathway for PTA to PG and 6-MP.

The second proposed pathway of PTA metabolism to 6-MP is the direct formation of 6-MP from PTA. This reaction is dependent on GSH and occurs to some extent nonenzymatically, because 6-MP accumulation was observed in incubations containing no protein. A likely mechanism for the nonenzymatic formation of 6-MP that is consistent with the effect of pH on the rate of the reaction is shown in Fig. 6B. At low pH, the acrylic acid moiety of PTA is protonated, thus increasing the electrophilic character of the -carbon. A Michael addition of GSH to the electrophilic -carbon in PTA results in the formation of a PTA-GSH conjugate from which 6-MP is subsequently released (Fig. 6B). Deprotonation of the carboxylic acid moiety with increasing pH markedly decreases the electrophilicity of the -carbon, thus decreasing the rate of the reaction. The residual GSH-acrylate conjugate formed in the reaction was not detected by our methods, possibly because of its instability and/or low absorptivity.

Our results suggest that GSTs can catalyze the formation of 6-MP from PTA. Upon the addition of liver homogenate, a 2-fold increase in 6-MP accumulation compared with buffer-only incubations was observed. This increase is unlikely to be due to metabolism of PG to 6-MP similar to that observed in kidney, because expression of -GT is very low in liver (Monks and Lau, 1987). The absence of metabolism of PG to 6-MP in liver was further confirmed by the observation that neither acivicin nor AOAA had any effect on PG or 6-MP accumulation in liver homogenate incubations. Rather, the increase in 6-MP formation is likely to be due to GST-mediated metabolism of PTA. Further support for this hypothesis was provided by the finding that time-dependent accumulation of 6-MP was faster in incubations containing purified rat liver or recombinant human GSTs than in buffer-only incubations.

It is of considerable interest to compare the structure and metabolism of the 6-MP prodrug azathioprine (Fig. 1A) to that of PTA. The nitro-conjugated double bond of the imidazole ring of azathioprine is a Michael acceptor similar to the acrylic acid moiety of PTA. Azathioprine is cleaved in vitro to 6-MP nonenzymatically by a nucleophilic attack of sulfhydryl groups, primarily GSH, on the -carbon in the activated double bond. The rate of this reaction increases with increasing pH and increasing GSH concentration due to increased efficiency of nucleophilic attack by GSH on the -carbon (Chalmers et al., 1967). Further studies showed that GSTs could catalyze the formation of 6-MP from azathioprine (Kaplowitz, 1976). The in vitro metabolism of azathioprine appears therefore to be analogous to the direct metabolism of PTA to 6-MP (Fig. 6B). The difference in reactivity with pH between azathioprine and PTA can be explained by the fact that the acrylic acid moiety of PTA is an ionizable group that loses its Michael acceptor characteristics with increasing pH, whereas the Michael acceptor of azathioprine is largely unaffected by changes in pH. In vivo studies using [³⁵S]azathioprine showed that the radiolabeled sulfur was mostly excreted as 6-MP and its metabolites, indicating that the in vivo cleavage of azathioprine is similar to that observed in vitro. However, radiolabeled [³⁵S]thioimidazolyl metabolites were also detected, indicating that azathioprine had been cleaved between the sulfur and the C-6 carbon of the purine ring (de Miranda et al., 1973). This pathway of azathioprine metabolism is strikingly similar to the conversion of PTA to PG (Fig. 6A).

The acquired multidrug resistance of tumor cells poses a serious problem in cancer chemotherapy. A large body of evidence now suggests that changes in tissue GST expression, elevation of GST activity and GSH content, detected in several drug-resistant tumor cell lines, are among the mechanisms involved in the drug resistance (Batist et al., 1986; Ahn et al., 1994; Chen and Waxman, 1995; Hayes and Pulford, 1995; Gulick and Fahl, 1995). GSTs are therefore promising candidates for selective targeting of anticancer agents to tumor cells. Recently, GSH analogs were designed and shown to be selectively metabolized by GSTs to yield a cytotoxic alkylating agent (Lyttle et al., 1994). Among these analogs, TER286 was metabolized to the active cytotoxic drug by GST P1-1, the most commonly up-regulated isozyme in tumor cells, and by GST A1-1, which is frequently overexpressed in cells with acquired resistance to nitrogen mustards (Tew, 1994; Morgan et al., 1998). Encouraging data on TER286 cytotoxicity have led to its consideration as a clinical candidate (Morgan et al., 1998). In light of these findings, it is of considerable interest to find that rat liver GSTs and human recombinant GSTs can metabolize PTA to 6-MP in a GSH-dependent manner. Comparison of the relative catalytic efficacies of the human recombinant isozymes revealed that GST A1-1 was the most effective isozyme, whereas GST P1-1 and GST M1-1 were less but equally effective in catalyzing the reaction. These findings suggest that PTA may function as a 6-MP prodrug, targeting tumor cells expressing high levels of GST and GSH. Although both TER286 and PTA are GST-activated prodrugs, the approach described in this manuscript is distinct from that described by Lyttle et al. (1994) and Morgan et al. (1998), in that PTA acts as a typical GST substrate, whereas TER286 apparently competes with GSH for its binding site on these enzymes.

PTA may also be metabolized to 6-MP by tumor cells through PG formation and further metabolism by -GT, dipeptidases and -lyases, since many tumor cells express high levels of -GT (Magnan et al., 1982). As PG has previously been shown to be metabolized to 6-MP selectively in the kidney (Hwang and Elfarra, 1993; Elfarra and Hwang, 1993), the discovery that formation of PG is the predominant pathway in the in vitro metabolism of PTA suggests that PTA may also be a kidney selective prodrug of 6-MP.

In vivo experiments in rats showed no liver or kidney toxicity due to PTA administration at the dose used (100 mg/kg). However, only a small amount of the PTA dose administered was recovered as 6-MP and its further metabolite, TU. Further experiments are needed to investigate the effect of dose variation on PTA metabolism and toxicity and to determine the optimum vehicle and route of administration for PTA.